AER VB ed (Wy MPD, 31-39 ECONOMICS OF INFORMATION Where Are We in the Theory of Information? By J. Himsuterren* As the “knowledge industry” booms in the world of affairs (see F. Machlup 1962 and J. Marschak 1968), the economics of information has been blooming with strik- ing and novel ideas in the intellectual realm. The rate of advance is suggested by the remarkable number of papers I will be citing that have not yet seen formal pub- lication. To keep the topic within bounds, I must cut drastically. First, I limit myself to theoretical innovations. Even so, many topics, such as the following, must be omitted. (1) The informational presupposi- tions underlying analytical concepts like the demand curve, short-run versus long- run cost functions, and imperfect competi tion. (2) The crucial role of information in bargaining and game theory. (3) Keynesian disequilibrium, in macro theory, asan infor- mational disfunction of the decentralized market economy. (4) Interpretations of unemployment as specialization in search for better opportunities. (5) Money as an institution economizing on knowledge that would otherwise be necessary to com- plete transaction chains. (6) The prospect of emergent information as determinant of demand for “liquidity” and of speculative behavior. (7) Adaptive expectations and * Professor of economics, University of California, ‘Los Angeles. Research on this topic was supported in part by the Social Science Research Council. Thanks for suggestions and ideas are due to Michael Darby, Harold Demsetz, Edward Gallick, and Axel Leijon- hufvad, au other learning models. (8) Efficient flows of information within multi-person organi- zations. I must set aside fascinating and important developments in these areas to concentrate upon my central theme—the microeconomics of information proper— or somewhat more explicitly: the produc- tion, dissemination, and manipulation of information in a market context. ‘The microeconomics of information in this narrower sense is an outgrowth of the economic theory of uncertainty. Uncer- tainty is summarized by the dispersion of individuals’ subjective probability (or belief) distributions over possible states of the world. Information, for our purposes, consists of events tending to change these probability distributions. A rather differ- ent concept of “information” is employed in communications and statistical theory, according to which a dispersed probability distribution is called less “informative” than a concentrated one (for certain ap- plications, see H. Theil). This latter con- cept uses the term “information” merely as a negative measure of uncertainty. But it is changes in belief distributions—a pro- cess, not a condition—that constitute here the essence of information. Note that the economics of information is active where the economics of uncertainty is passive. Mere adaptation to a given state of ignorance via optimal “terminal” action is still in the realm of the economics of un- certainty; in going beyond this to consider the alternative of gathering more evidence32 AMERICAN ECONOMIC ASSOCIATION prior to terminal action, we enter the domain of the economics of information. I. Information-Involved Behaviors; Categories of Information The distinction between passive and ac- tive adaptations to the fact of uncertainty isa familiar one in the theory of statistical decision, which falls naturally into the two divisions of: (a) criteria for action on given sample evidence, versus (b) design of ex- periments for the generation of additional evidence. But where the statistician con- siders only acquisition of information by experiment, we want to examine the wider opportunities that arise for acquiring and using information in a market context. Table 1 classifies behavior modes for possessors and for seekers of economically valuable information. The possessor can in general benefit simply by private use of the information for his own productive or consumptive decisions. But in a market context it might also be possible for him to profit from sale of the information to others. The information-seeker might cor- respondingly find it advantageous to pro- duce socially “new” information by direct inquiry of Nature (research) or to purchase “secondhand” information in the market. Viewed as a tradeable commodity, in- formation has (as we shall see) a number of special features. The most novel aspects of the economics of information stem, how- ever, from the behavior possibilities indi- cated by the third line of Table 1. In the market process information can be re- garded as “pulled” from the possessor by Tame 11 Possessor of Information Private Use Sale Gratuitous dissemination (“pushing”) . Deception—authentication MAY 1973 purchase, i.e., by payment of an explicit price. But what is surprising, the possessor may find it preferable to give away this valuable commodity, to disseminate it without pull of compensation. Indeed it may be highly profitable for him to incur costs so as to gratuitously “push” informa- tion to potential recipients! As for the in- formation-secker, his knowing that the ossessors are so motivated may lead to adoption of a monitoring or listening mode of learning behavior (J. Marshall). Standing somewhat apart in the Table, but a crucial element whenever informa- tion is to be disseminated (whether gratu- itously or by sale), is the possibility of deception—leading to the counter-activities ‘of evaluation on the part of the informa- tion-seeker and authentication on the part of the disseminator. ‘Tamue 2—Economtcatty 81 ‘Inronwation Arran |. Certainty Diffusion Applicability: particular vs. general |. Content: Environmental vs. behavioral “Tastes, endowments, technology vs. market parameters 5. Decision-relevance Listed in Table 2 are a number of at- tributes affecting the value of information to potential users or producers. Cerlainty refers to the degree of concentration of posterior belief distributions dictated by the information; fully certain information assigns 100 percent probability.to a single value of the variable being predicted. The {PORMATION-INVOLVED Mopes OF BEHAVIOR Seeker of Information 1. Production (research) 2. Purchase (“pulling”) 3. Monitoring 4, EvaluationVOL. 63 NO. 2 extent of Diffusion will obviously affect the scarcity value of information. A pplicabil ity is, at one extreme, particular to a single economic agent—as when I learn of oil under my land. A new process for extract- ing oil more cheaply from everyone’s land would be of general applicability. The Con- tent of information may be subclassified i a variety of ways. One distinction com- monly made is between information about the physical environment versus informa- tion about the strategies or behavior of other individuals; this is relevant for con- siderations of Pareto optimality (R. Rad- ner). Another useful classification runs in terms of the elements of choice-theoretic structures in economics: tastes, endow- ments (resources), technology (production functions), and market characteristics (price or quality of traded goods). The as- pect of Decision-relevance has been brought forward (Marschak 1964, Marschak and K. Miyasawa) as a corrective to attempts to quantify economically relevant informa- tion by the “bit” measure of communica- tions theory. Time and space fortunately preclude systematic coverage of the full range of behaviors in Table 1 over the information attributes in Table 2. As it happens, the existing literature is divided into two dis- inet branches. In the first, individuals are assumed subject to technological uncer- tainty only. They are unsure about, and therefore interested in information con- cerning, only their resource endowments and/or productive opportunities. Analysts here generally have assumed perfect mar- kets (except, possibly, for the market for information itself)—so that an equilibrium integrating all supply-demand offers is instantaneously and costlessly attained. In the second division of the literature, technological uncertainty is assumed away ; there is market uncertainty instead. Each individual is supposed to be fully certain about his own endowment and productive ECONOMICS OF INFORMATION B opportunities, but only imperiectly in- formed about the supply-demand offers of others, The two areas are reviewed in the sections that follow. II. Technological Information: The Underinvestment Issue A. General Information and Patents Patents permit the conversion of certain types of information (ordinarily, produc- tion-function information of general ap- plicability) into legally recognized prop- erty. The traditional position has been that while patent royalty fees hinder the optimal utilization of information once produced, lack of appropriability would otherwise lead to underinvestment in new technological ideas (Machlup, 1968). K. J Arrow (1962) has developed this line of thought further, maintaining that there would be underproduction of ideas even with a patent system because: (1) inven- tion is risky and all risky activities are underexploited for lack of complete condi jonal-contract markets; (2) appropriabil- ity is imperfect, since patent protection is only partially effective; and (3) royalty schemes do not generally capture all the benefits for the inventor. This seems quite convincing, but recent contributions show that there are also con- siderations cutting in the opposite direc- tion. Y. Barzel has argued that undis- covered ideas are like fish in the sea, sub- ject to the rule of capture. Since the patent right goes to the first in possession, with perfect patents competitive invention would be biased toward prematurity. The rule of capture leads to too many too small fish being caught! On quite another ground, J. Hirshleifer has shown that the standard analysis entirely ignores the profit possibilities implicit in the bottom line of Table 1—the “pushing” of informa- tion. The inventor, first in the know, might be in a position to predict and therefore speculate upon price revalua-ue AMERICAN ECONOMIC ASSOCIATION tions ensuing from the publicizing of his information. (The Hall process, for ex- ample, increased the value of bauxite ores.) Note that the profit opportunities here dictate the widest dissemination of the information; the speculative “pushing” motive, in contrast with the sale motive that the patent institution facilitates, furthers both the utilization and the pro- duction of information. As an additional point, where there are differences of belief, incentives exist for cooperative invest- ment in information acquisition that may easily exceed the social value thereof. B. Particular Information and the Disclosure Problem F. A. Hayek’s pioneering 1945 article emphasized the importance of “knowledge of particular circumstances of time and place” (information of particular applica- bility) as opposed to “scientific know!- edge” (information of general applicabil- ity). Hayek went on to argue that a cen- tralized economy would find it difficult to communicate particular information to decision-making points—whereas in a mar- ket economy all the relevant aspects of such information are efficiently dissemi- nated via the price system. Interestingly, Hayek speaks only of the use of information—not its production. While it is plausible to argue that in- dividuals are appropriately motivated in a market economy to generate self-regarding information, other parties may have a com- parative advantage in its actual produc- tion, But the motivation for outsiders to produce particular information about others for sale (e.g., if I have a way of find- ing out whether there is oil under your land) is impaired by the unavoidable mo- nopsony on the buying side. And attempts to profit from “pushing” the information will not work very well, since trying to take the requisite speculative position (buy up the undervalued property) will MAY 1973 signal the content of the information. ‘Matters appear in a somewhat different light, however, if we consider information particular not to atomic individuals but to publicly held corporations. Here the dis persed universe of actual and potential owners of securities constitutes an im- personal market for sale of information about particular firms. And indeed security analysis is a thriving business. Further- more, the securities markets themselves provide a particularly efficient medium for achieving speculative gains from the acquisition and subsequent dissemination of particular information. Information generation about particular firms has been studied by E. F. Famaand A.B. Laffer. They argue that there will be waste of resources in information produc- tion motivated by the prospect of specula- tive trading in the corporation’s stock (or, at one remove, by the prospect of sale to others who will use the information for that purpose). For the trading will be merely redistributive in impact. The i formation may also be of value for im- proving productive allocations, but there will remain a private motivation for ex- pending resources beyond the level war- ranted by anticipated productive improve- ments. (But it should be recalled that without the speculative motive the in- trinsic monopsony on the purchasing side would deter efficient outside producers of particular information, so the net balance remains unclear.) The potential conflict of interest between corporate owners and managers has im- portant implications for the acquisition and dissemination of information. H. G. Manne argues that managerial (“insider”) trading in the corporation’s stock provides socially useful compensation for entrepre- neurial innovation and concludes that “disclosure laws” limiting such trading are unwise. But trading profits stem from superior information, whether about favor-VOL. 63 NO. 2 able or unfavorable developments. Thus, insider trading can reward unexpectedly bad managerial performance just as hand- somely as good performance; all the insider need do is sell short on the basis of prior knowledge of his own mistakes. Second, the previously discussed overinvestment in in- formation comes up once more—in es- pecially strong form. Insiders are by def- inition well placed to secure information and may even be able to use corporate funds (rather than their own) to cover the costs of doing so. As an alternative to higher’ salaries, then, this mode of re- muneration of management entails serious inefficiencies. C. Transferability of Information One of the key themes of the analysis to this point is that if information were per- fectly transferable (cither by sale or by pushing), it would be “over-produced” (but see H. Demsetz). The reason is that changes in probability beliefs (and there- fore in market prices) lead to wealth re- distributions as well as to productive- consumptive adaptations. The sum of gains from the adaptations represents the social value of the information and is the maximum that can efficiently be paid to cover the costs of producing it. But while the net social value of the redistributions is zero, individuals would obviously be will- ing to pay to be on the winning side of the shufile. The sum of the two classes of po- tential payments would overcompensate the information producer. Limited transferability is a countervail- ing consideration. The first problem is that of authenticity: a seller always claims to be telling the truth, but how is a buyer to know? Sometimes authenticity is mani- fest in the information itself (‘“Behold!”), or can be made manifest at some cost. Alternatively, authenticity may be con- veyed by guarantees or by other tech- niques to be discussed below. Authenticity ECONOMICS OF INFORMATION 3s being assumed, a number of other elements entering into marketability have been considered by Y. Noguchi. Even with patent protection, it must be possible to detect unauthorized use and to identify the user. Unpatented information is safe- guarded by secrecy, which is always com- promised by sale. The key problem for the existence of a market in such information is the prevention of unauthorized resale. Fama and Laffer, in their paper, assumed completely effective protection against such resale; Noguchi, on the other hand, argues that resale prohibition can never be effective. The truth, naturally, is some- where in between. IIL Market Information The literature dealing with information about market parameters—more specifi- cally, information about the terms on which potential trading partners are will- ing to do business—assumes away any un- certainty about technological and other exogenous features of the economic prob- lem. Information is not wanted here to provide a better basis for the individual’s own supply-demand offers, but rather solely to permit taking advantage of the offers of others. Nevertheless, the logical categories of Tables 1 and 2 apply. Infor- mation about trading partners, like tech- nological information, can be produced or sold or “pushed.” And the difficulty of marketing or otherwise disseminating trad- ing partner information remains essentially the same as for technological information. ‘The analysis of market information took a “great leap forward” with the path- breaking article of our Chairman G. Stigler which spelled out and partially solved some of the major questions under this heading, including: (1) the nature and extent of search and advertising behavior (the latter an instance of “pushing” in- formation), and (2) equilibrium in a mar- ket with continuing search and advertis-36 AMERICAN ECONOMIC ASSOCIATION ing. Search and advertising are comple- mentary informational processes. The searcher locates specific offers; the adver- tiser “pushes” the fact of his existence and, possibly, some details about his terms for dealing. For brevity, I assume in this discussion that buyers search while sellers (or some sellers) advertise—though in gen- eral there will be some of each activity on either side of the market. ‘A. Information about Price In the analysis of price information, as opposed to quality or brand information, it is generally assumed that authenticity is manifest—even though deceptive ways of quoting price are not unknown in the world of affairs. This simplification, to- gether with the measurability of price, has permitted rigorous mathematical solu- tions of some outstanding questions. Rational price search behavior (mathe- matically, the optimum stopping rule) has been analyzed by a number of authors— including J. J. McCall, J. L. Gastwirth, and L. G. Telser. Stigler had originally assumed that the searcher would investi- gate a sample of predetermined size. The later authors showed that a sequential process is generally superior; the optimal policy is to accept an offer if the terms are superior to a predetermined reservation price. The reservation price itself may be a function of the earlier observations, if these are informative (i.., if the distribu- tion of price offers is not known 4 priori). ‘And if the searching takes time, the reser- vation strategy should take account of the fact that both the returns and the op- portunity costs of search are influenced by temporal factors (A. A. Alchian and R. Gronau). The returns from search are also affected by the importance of the com- modity in the budget and by the durability of the information gained. The costs of search depend upon the distance of sellers from the buyer and from each other (i.e., MAY 1973 their locational concentration) and the extent to which they have been identified and their offers made visible by advertis- ing or other means. ‘The seller’s advertising decision is bound up with his reciprocal anticipations of buyer search behavior, his standing in the price distribution (assuming he knows it), and the advertising effort of competi- tors. The distinctiveness of the commodity, the rarity of buyers, and the efficiency of the communications media are all in- volved. Finally, middlemen (brokers) may emerge who specialize in the acquisition of market information and its sale to traders on one or both sides of the market. ‘The second major problem under this heading is market equilibrium and its consistency with continuing search/adver- tising activity. The difficulty is that search and advertising appear to lead toward a unique price equilibrium rather than a distribution of prices. Stubbornly high- priced sellers will lose customers over time, and low-priced sellers will become swamped with clients—leading in both cases to a corrective movement of price offers. As this correction takes place the gains from search and advertising dwindle. So these processes seem to be self-limiting, con- sistent with transition toward but not achievement of equilibrium. Of course, the exogenous conditions de- termining demand and supply are ever- changing; as Stigler points out, change maintains price dispersion and, therefore, the rationale for search and advertising. This amounts to saying that we are always in transit to, never at, equilibrium. But if the exogenous changes can be regarded as drawings from a fixed probability distribu- tion, the economic system may converge toward a statistical equilibrium containing price dispersion and associated search/ advertising behavior (J. R. Green). Stigler’s original paper also mentions, however, endogenous factors that tend toVOL. 63 NO. 2 preserve price dispersion—mobility and forgetting. Mobility of buyers and sellers into and out of the market (for example, in an intergenerational life cycle model) will provide a continuing demand for the informational processes of search and ad- vertising. Forgetting achieves the same result by destroying information; forget- ting is akin to a mutation probability off- setting the adaptive directive effect of search. Finally, M. Rothschild and M. Yaari (reported in Rothschild) have developed a model in which sellers engage in experi- mental price variation while buyers are simultaneously searching. The combina- tion results in sellers and buyers never be- coming perfectly informed, so that price dispersion and search behavior persist. B. Information about Quality Uncertainty about quality poses an in- trinsically more difficult problem than un- certainty about price. There are two main reasons. First, quality may be multi- dimensional, unquantifiable in some re- spects, and may contain an irreducible sub- jective element. Second, the authenticity of the claims made by sellers now becomes a most serious question for market par- ticipants. In the typical situation in the literature, information endowments are asymmetrical: specifically, let us assume that the seller knows the quality of the product but the buyer does not. G. A. Akerlof has shown that the informational asymmetry may lead to what is called “adverse selection” in insurance jargon. Suppose that buyers can only judge quality by the average level in the market. Then sellers with inferior products are encouraged to offer them for sale—while those with superior products are correspondingly discouraged. Further- more, there is a parallel and reinforcing process that corresponds to “moral haz- ard” in insurance jargon (Arrow 1963); ECONOMICS OF INFORMATION ar this refers to the temptation to sellers to deliberately degrade the quality of the product in response to buyer ignorance. Countervailing institutions tend to emerge, establishing distinctiveness and responsibility through brand names. De- vices employed include informative ad- vertising, guarantees, “signalling,” and in- dependent information producers. Inform- ative advertising, an elusive though per- haps not so rare category, convinces by its content—eg., by citing known or verifi- able facts and drawing valid inferences from arguments. Guarantees lend convic- tion because the seller makes himself vulnerable to penalty should the claim prove incorrect. “Signalling,” a concept introduced by M. Spence, is a kind of implicit guarant the seller engages in some ancillary ac- tivity that would be irrational were his claims not correct. In the example used by Spence, higher quality workers signal by acquiring education (even if education does not contribute to productivity), on the hypothesis that employers know that higher quality workers can more easily or cheaply undergo the educational ordeal. Independent information producers (in- cluding both for-profit concerns and not- for-profit certifying agencies like medical associations) have the same problem of establishing their own brand names before they can be of value to others. Buyers, of course, will not be entirely incapable of evaluating quality for them- selves. P. Nelson (1970) distinguishes two types of consumer investigative behavior: inspection (Nelson uses the less appropriate designation “search”) is evaluation that can take place without purchase, experi- ence only after purchase. One or the other process will in any given situation be the cheaper, and commodities can accord- ingly be classed into inspection goods and experience goods. Nelson maintains that a larger number of brands will tend to be8 AMERICAN ECONOMIC ASSOCIATION sampled for high-purchase-frequency items and also for inspection goods as against experience goods. Then inspection goods should tend to be more competitively mar- keted—monopoly power will be less if consumers sample widely. Elasticity of demand approaches zero during the experi- mental phase when investigating by ex- perience, since the consumer cannot pre- sume in advance that a higher priced brand is not correspondingly higher quality; this point does not apply to inspection goods, which are not actually purchased in the experimental phase. ‘Nelson considered the market implica- tions of quality evaluations by buyers. M. R. Darby and E. Karni have examined the effect of buyers’ difficulty in evaluat- ing quality upon seller behavior. Unlike Akerlof, they consider a situation in which sellers accept responsibility for quality claims. The temptation to depreciate quality that constituted “moral hazard” for Akerlof becomes “fraud” for Darby and Karni. As in Akerlof’s situation, countervailing market arrangements tend to emerge. They include service contracts (the seller in effect commits himself to maintaining a level of service from the good), the client relationship, and of course the implicit guarantee of a brand name reputation. Note that we have here another illustration of the general problem of authenticating disseminated informa- tion. C. Market-Information Processes and Social Efficiency It seems plausible, and several authors maintain, that there are favorable ex- ternalities in the discovery and dissemina- tion of valid market information and, hence, that these activities should be en- couraged by public policy. Akerlof has pointed to such an externality where markets are threatened by information dis- parities (the adverse-selection and moral- MAY 1973 hazard problems). Darby and Karni point to the social waste of resources devoted to fraud (while warning that government “protective” activities cannot be assumed to be disinterested either). Arrow (1963) had suggested earlier that there would be underproduction of valid market informa- tion, a contention that parallels his argu- ments about technological information (1962). On the other hand, in the model of Spence even the resources devoted to sig- nalling perfectly authentic information are socially unproductive (although privately remunerative); as in the corresponding contentions of Hirshleifer and Fama and Laffer for technological information, the gains from better information can some- times be merely redistributive. What about advertising regarded as an information-transfer process? I must limit myself to a few points here. First, even short of what Darby and Karni would call fraud, it is evident that resources are being wasted conveying inauthentic informa- tion, Nelson (1972), on the other hand, emphasizes consumers’ power to police the informative content of advertising, par- ticularly in the case of inspection goods. Even for experience goods, Nelson argues, the fact of advertising itself conveys an assurance of quality. The higher quality brand will, other things equal, have a comparative advantage in acquiring more customers by advertising—since it will re- tain a larger fraction of them on repeat sales. Nelson’s argument, in effect, is that advertising is a “‘signal” in Spence’s sense. Finally, because of our limited mental- processing capacities there may well be a congestion effect at work: if, as G. A. Miller says, we can only keep in mind seven things at once (plus or minus two), one good datum may just be driving out another. REFERENCES G. A. Akerlof, “The Market for ‘Lemons’:VOL. 63 NO. 2 Quality Uncertainty and the Market Mecha- nism,” Quart. J. Econ., Aug. 1970, 84, 488- 500. A. A. Alchian, “Information Costs, Pricing, ‘and Resource Unemployment,” West. Econ. J, June 1969, 7, 109-28. K. J. Arrow, “Economic Welfare and the Allo- ation of Resources to Invention,” in The Rate and Direction of Inventive Activity: Economic and Social Factors, Universities— Nat. Bur. Econ. Res. Conference series, Princeton 1962. —— “Uncertainty and the Welfare Eco- nomics of Medical Care,” Amer. Econ. Rev., Dec. 1963, 53, 941-73. Y. Bareel, “Optimal Timing of Innovations,” Rev. Econ. Statist., Aug. 1968, 50, 348-55. M.R. Darby, and E. Karn, “Free Competition and the Optimal Amount of Fraud,” Ohio State Univ. Dept. of Economics, Div. for Econ. Res. Report 7227, May 1972. 1H. Demsetz, “Information and Efficiency: An- other Viewpoint,” J. Law Econ., Apr. 1969, 12, 1-22, E. F. Fama and A. B. Laffer, “Information and Capital Markets,” J. Bus., July 1971, 44, 289-98. : J. L. Gastwirth, “On Probabilistic Models of Consumer Search for Information,” Unpub- lished paper, Johns Hopkins Univ. and the Nat. Sci. Found., 1971 J. R. Green, “The Nature and Existence of Stochastic Equilibria,” Harvard Inst. of Econ. Res. Discussion Paper No. 193, 1971. R. Gronau, “Information and Frictional Un- employment,” Amer. Econ. Rev., June 1971, 61, 290-301. F. A. Hayek, “The Use of Knowledge in So- ciety,” Amer. Econ. Rev., Sept. 1945, 35, 519-30. J. Hirshleifer, “The Private and Social Velue of Information and the Reward to Inven- tive Activity,” Amer. Econ. Rev, Sept. 1971, 61, 561-74. F. Machlup, The Production and Distribution of Knowledge in the United States, Prince- ton 1962. , “Patents,” International Encyclopedia of the Social Sciences, New York 1968, 11, 461-72. ECONOMICS OF INFORMATION o H. G. Manne, Insider Trading and the Stock ‘Market, New York 1966. J. Marschak, “Economies of Inquiring, Com- municating, Deciding,” Amer. Econ. Rev., Proc., May 1968, 58, 1-18. “Problems in Information Economics,” in C. P. Bonini, R. K. Jaedicke, and H. M. Wagner, eds., Management Controls: New Directions in Basic Research, New York 1964, and K. Miyasawa, “Economic Com- parability of Information Systems,” Jnt. Econ, Rev., June 1968, 9, 137-74. J. Marshall, “Private Incentives and Public In- formation,” Univ. of California at Santa Barbara, Dept. of Econ., Working Paper in Economics No. 14, Oct. 1972. J. J. McCall, “The Economics of Information and Optimal Stopping Rules,” J. Bus., July 1965, 38, 300-17. G. A. Miller, “The Magical Number Seven, Plus or Minus Two: Some Limits on our Capacity for Processing Information,” Psychol. Rev., 1956, 63, 81-97. P. Nelson, “Information and Consumer Be- havior,” J. Polit. Econ., Mar—Apr. 1970, 78, 311-29. ——\, “Advertising as Information,” Paper presented at Nat. Bur. Econ. Res—Nat. Sci. Found. Conference on Economics of In- formation, April 7-8, 1972. Y. Noguchi, An Economic Theory of Diffusion of Knowledge, Unpublished doctoral disser- tation, Yale Univ. 1972, R. Radner, “Competitive Equilibrium Under Uncertainty,” Econometrica, Jan. 1968, 36, 31-58. 'M. Rothschild, “Models of Market Organiza- tion with Imperfect Information: A Survey,” Harvard Inst. of Econ. Res. Discussion Pa- per No. 224, Dec. 1971. M. Spence, “Market Signalling,” Harvard Univ. Kennedy School Discussion Paper No. 4, Feb. 1972. G. Stigler, “The Economics of Information,” J. Polit. Econ., June 1961, 69, 213-25. L. G, Telser, “Searching for the Lowest Price,” Amer. Econ, Rev., Proc., May 1973. H. Theil, Economics and Information Theory, Chicago 1967.